Hi Rutger, sorry for the delay.

I would go with creating cope3 at first level and feed this up, for
similar reasons for using a paired t-test instead of unpaired, when
appropriate. Hope this makes sense! Sorry you have to wait so long for
such a short answer!

Thanks, Steve.



On Fri, 19 Sep 2003, Goekoop, R. wrote:

> Hi Steve,
>
> Thanks - it makes sense, but I probably took a narrow turn. I'll rephrase
> the problem. I want to examine the effect of medication on a particular
> contrast between two conditions, say A > B. At lower level, PEs are
> calculated for both A and B (say pe1 and pe2), and these are identical to
> cope1 and cope2 in this case (I think, since they look identical in AFNI).
> Now, it is possible to calculate an additional COPE image on this (first)
> level describing A > B using pe1 and pe2. This will be cope3 (at first
> level). I could then use all cope3 images from different subjects and
> examine the effect of medication on these (first level) activation maps in a
> higher level analysis. All set.
>
> However, it might also be possible to feed the lower level cope1 and cope2
> images (same as PEs of A and B) into a higher level analysis, calculate a
> contrast A > B on this level (say /blah.gfeat/cope1/stats/cope1.img etc),
> and examine the effect of medication on these (second level) activation maps
> in a third level analysis. Would there be any significant difference between
> both approaches and if so, which one would you prefer, perhaps considering
> the underlying statistics?
>
> Thanks a billion,
>
> Rutger.
>
> -----Original Message-----
> From: Stephen Smith [mailto:[log in to unmask]]
> Sent: vrijdag 19 september 2003 16:38
> To: [log in to unmask]
> Subject: Re: [FSL] FEAT: Choosing the level of statistical analysis
>
>
> Hi Rutger,
>
> Each voxel at first level has a different cope and variance of cope
> (varcope), and the higher-level analysis using FLAME is designed to get the
> best possible analysis using the low-level varcopes, separately for each
> voxel. Therefore I suspect that it is preferable to to a normal
> first- and then second-level (and then third if relevant) analysis and then
> do roi analysis on the highest-level output.
>
> I hope I've understood your question correctly. Does this make sense?
>
> Cheers, Steve.
>
>
>
> On Thu, 18 Sep 2003, Goekoop, R. wrote:
>
> > Dear FSL-users,
> >
> > In analysing data from a study that examined the effects of medication
> > challenge on brain activation patterns, it is possible to calculate
> > most COPE images of interest (COIs :-) using basic COPE images at the
> > first level and examine the effects of medication on these COIs using
> > a second level analysis. It is however also possible to enter basic
> > COPEs as inputs to the second level analysis, calculate the COIs on
> > this level, and then use a third level analysis to extract the final
> > effect of medication from these higher-level COIs. Would there perhaps
> > be a reason why any of these two approaches could be considered
> > optimal (despite the carrying up of the lower-level variances)? It
> > probably depends a lot on study design and the kind of (basic) COPE
> > images that are chosen as inputs to the higher level analyses, but do
> > the different statistical approaches used on the various levels have
> > any bearing on this issue?
> >
> > Thanks,
> >
> > Rutger.
> >
> > Drs. R. Goekoop, MD.
> > Department of Neurology
> > Vrije Universiteit Medical Centre
> > P.O. Box 7057, 1007 MB
> > Amsterdam, the Netherlands
> > Phone: +31 20 444 0316
> > E-mail:  <mailto:[log in to unmask]> [log in to unmask]
> >
> >
> >
>
>  Stephen M. Smith  MA DPhil CEng MIEE
>  Associate Director, FMRIB and Analysis Research Coordinator
>
>  Oxford University Centre for Functional MRI of the Brain
>  John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
>  +44 (0) 1865 222726  (fax 222717)
>
>  [log in to unmask]  http://www.fmrib.ox.ac.uk/~steve
>

 Stephen M. Smith  MA DPhil CEng MIEE
 Associate Director, FMRIB and Analysis Research Coordinator

 Oxford University Centre for Functional MRI of the Brain
 John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
 +44 (0) 1865 222726  (fax 222717)

 [log in to unmask]  http://www.fmrib.ox.ac.uk/~steve